
import matplotlib.pyplot as plt
import pandas as pd
import statsmodels.api as sm
from sklearn.preprocessing import OrdinalEncoder
import seaborn as sns
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
import numpy as np
from scipy import stats
import matplotlib.style as style


# Read all the necessary csv files into individual dataframes
df_2000 = pd.read_csv('./data/atp_matches_2000.csv')
df_2001 = pd.read_csv('./data/atp_matches_2001.csv')
df_2002 = pd.read_csv('./data/atp_matches_2002.csv')
df_2003 = pd.read_csv('./data/atp_matches_2003.csv')
df_2004 = pd.read_csv('./data/atp_matches_2004.csv')
df_2005 = pd.read_csv('./data/atp_matches_2005.csv')
df_2006 = pd.read_csv('./data/atp_matches_2006.csv')
df_2007 = pd.read_csv('./data/atp_matches_2007.csv')
df_2008 = pd.read_csv('./data/atp_matches_2008.csv')
df_2009 = pd.read_csv('./data/atp_matches_2009.csv')
df_2010 = pd.read_csv('./data/atp_matches_2010.csv')
df_2011 = pd.read_csv('./data/atp_matches_2011.csv')
df_2012 = pd.read_csv('./data/atp_matches_2012.csv')
df_2013 = pd.read_csv('./data/atp_matches_2013.csv')
df_2014 = pd.read_csv('./data/atp_matches_2014.csv')
df_2015 = pd.read_csv('./data/atp_matches_2015.csv')


df_2016 = pd.read_csv('./data/atp_matches_2016.csv')
df_2016


df_2017 = pd.read_csv('./data/atp_matches_2017.csv', index_col=False)
df_2017

/var/folders/gx/9ss39m455rl9klmbs793xqnm0000gn/T/ipykernel_27094/855321516.py:1: ParserWarning: Length of header or names does not match length of data. This leads to a loss of data with index_col=False.
  df_2017 = pd.read_csv('./data/atp_matches_2017.csv', index_col=False)


df_list = [df_2000, df_2001, df_2002, df_2003, df_2004, df_2005, df_2006, df_2007, df_2008, df_2009, df_2010, df_2011, df_2012, df_2013, df_2014, df_2015, df_2016, df_2017]
df = pd.concat(df_list)
df


# Find out what % of our dataframe is missing (null)
df.isnull().sum().sum()/df.size

0.11131708101283858


# Output all the columns in the dataframe.
for c in df.columns:
    print(c)

tourney_id
tourney_name
surface
draw_size
tourney_level
tourney_date
match_num
winner_id
winner_seed
winner_entry
winner_name
winner_hand
winner_ht
winner_ioc
winner_age
winner_rank
winner_rank_points
loser_id
loser_seed
loser_entry
loser_name
loser_hand
loser_ht
loser_ioc
loser_age
loser_rank
loser_rank_points
score
best_of
round
minutes
w_ace
w_df
w_svpt
w_1stIn
w_1stWon
w_2ndWon
w_SvGms
w_bpSaved
w_bpFaced
l_ace
l_df
l_svpt
l_1stIn
l_1stWon
l_2ndWon
l_SvGms
l_bpSaved
l_bpFaced


# Drop all the unnecessary columns using indexes, and then display the dataframe
df = df.drop(df.columns[[0, 1, 3, 4, 5, 6, 7, 8, 9, 13, 16, 17, 18, 19, 23, 26, 27, 28, 29, 30]], axis=1)
df


# Now to check the columns we have:
for c in df.columns:
    print(c)

surface
winner_name
winner_hand
winner_ht
winner_age
winner_rank
loser_name
loser_hand
loser_ht
loser_age
loser_rank
w_ace
w_df
w_svpt
w_1stIn
w_1stWon
w_2ndWon
w_SvGms
w_bpSaved
w_bpFaced
l_ace
l_df
l_svpt
l_1stIn
l_1stWon
l_2ndWon
l_SvGms
l_bpSaved
l_bpFaced


df.isnull().sum().sum()/df.size

0.07592115909341068


df = df.dropna()


# Re-check how much of our data is missing after dropping all null entries.
print(df.isnull().sum().sum()/df.size)

0.0


df = df[(df['winner_hand'] == 'L') | (df['winner_hand'] == 'R') | (df['loser_hand'] == 'L') | (df['loser_hand'] == 'R')]


# Transform our data: Change the symbol associated with dominant hand of player
#R -> W_R / L_R and L -> W_L/ L_L

df.loc[df["winner_hand"] == "R", "winner_hand"] = 'W_R'
df.loc[df["winner_hand"] == "L", "winner_hand"] = 'W_L'
df.loc[df["loser_hand"] == "R", "loser_hand"] = 'L_R'
df.loc[df["loser_hand"] == "L", "loser_hand"] = 'L_L'
dummies_df = pd.get_dummies(df[['winner_hand', 'loser_hand']])
df_list = [df, dummies_df]
df = pd.concat(df_list, axis = 1)


#To decide how to label our surfaces, let's use a frequency chart of the surfaces present in the dataframe.
# Make a plot of all the surfaces.
pt = sns.countplot(x = df['surface'], data=None)


(df['loser_hand'] == 'U').count() 

#There is 1 person with U hand -> remove since it we just want R and L hand

df = df.loc[df["loser_hand"] != 'U']
df


dummy_df = pd.get_dummies(df['surface'])
df_list = [df, dummy_df]
df = pd.concat(df_list, axis=1)


df.dtypes

surface             object
winner_name         object
winner_hand         object
winner_ht          float64
winner_age         float64
winner_rank        float64
loser_name          object
loser_hand          object
loser_ht           float64
loser_age          float64
loser_rank         float64
w_ace              float64
w_df               float64
w_svpt             float64
w_1stIn            float64
w_1stWon           float64
w_2ndWon           float64
w_SvGms            float64
w_bpSaved          float64
w_bpFaced          float64
l_ace              float64
l_df               float64
l_svpt             float64
l_1stIn            float64
l_1stWon           float64
l_2ndWon           float64
l_SvGms            float64
l_bpSaved          float64
l_bpFaced          float64
winner_hand_W_L      uint8
winner_hand_W_R      uint8
loser_hand_L_L       uint8
loser_hand_L_R       uint8
loser_hand_U         uint8
Carpet               uint8
Clay                 uint8
Grass                uint8
Hard                 uint8
dtype: object


df.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 42923 entries, 0 to 384
Data columns (total 38 columns):
 #   Column           Non-Null Count  Dtype  
---  ------           --------------  -----  
 0   surface          42923 non-null  object 
 1   winner_name      42923 non-null  object 
 2   winner_hand      42923 non-null  object 
 3   winner_ht        42923 non-null  float64
 4   winner_age       42923 non-null  float64
 5   winner_rank      42923 non-null  float64
 6   loser_name       42923 non-null  object 
 7   loser_hand       42923 non-null  object 
 8   loser_ht         42923 non-null  float64
 9   loser_age        42923 non-null  float64
 10  loser_rank       42923 non-null  float64
 11  w_ace            42923 non-null  float64
 12  w_df             42923 non-null  float64
 13  w_svpt           42923 non-null  float64
 14  w_1stIn          42923 non-null  float64
 15  w_1stWon         42923 non-null  float64
 16  w_2ndWon         42923 non-null  float64
 17  w_SvGms          42923 non-null  float64
 18  w_bpSaved        42923 non-null  float64
 19  w_bpFaced        42923 non-null  float64
 20  l_ace            42923 non-null  float64
 21  l_df             42923 non-null  float64
 22  l_svpt           42923 non-null  float64
 23  l_1stIn          42923 non-null  float64
 24  l_1stWon         42923 non-null  float64
 25  l_2ndWon         42923 non-null  float64
 26  l_SvGms          42923 non-null  float64
 27  l_bpSaved        42923 non-null  float64
 28  l_bpFaced        42923 non-null  float64
 29  winner_hand_W_L  42923 non-null  uint8  
 30  winner_hand_W_R  42923 non-null  uint8  
 31  loser_hand_L_L   42923 non-null  uint8  
 32  loser_hand_L_R   42923 non-null  uint8  
 33  loser_hand_U     42923 non-null  uint8  
 34  Carpet           42923 non-null  uint8  
 35  Clay             42923 non-null  uint8  
 36  Grass            42923 non-null  uint8  
 37  Hard             42923 non-null  uint8  
dtypes: float64(24), object(5), uint8(9)
memory usage: 10.2+ MB


df = df.drop(df[['surface', 'winner_hand', 'loser_hand']], axis=1)


import matplotlib.pyplot as plt


df.columns

Index(['winner_name', 'winner_ht', 'winner_age', 'winner_rank', 'loser_name',
       'loser_ht', 'loser_age', 'loser_rank', 'w_ace', 'w_df', 'w_svpt',
       'w_1stIn', 'w_1stWon', 'w_2ndWon', 'w_SvGms', 'w_bpSaved', 'w_bpFaced',
       'l_ace', 'l_df', 'l_svpt', 'l_1stIn', 'l_1stWon', 'l_2ndWon', 'l_SvGms',
       'l_bpSaved', 'l_bpFaced', 'winner_hand_W_L', 'winner_hand_W_R',
       'loser_hand_L_L', 'loser_hand_L_R', 'loser_hand_U', 'Carpet', 'Clay',
       'Grass', 'Hard'],
      dtype='object')


import matplotlib.pyplot as plt
pt = df.hist(figsize=(20,20), xrot=45)
plt.show()


# df.dtypes
df = df.drop('winner_name', axis=1)
df = df.drop('loser_name', axis=1)


df = df.astype('float64')


df.dtypes

winner_ht          float64
winner_age         float64
winner_rank        float64
loser_ht           float64
loser_age          float64
loser_rank         float64
w_ace              float64
w_df               float64
w_svpt             float64
w_1stIn            float64
w_1stWon           float64
w_2ndWon           float64
w_SvGms            float64
w_bpSaved          float64
w_bpFaced          float64
l_ace              float64
l_df               float64
l_svpt             float64
l_1stIn            float64
l_1stWon           float64
l_2ndWon           float64
l_SvGms            float64
l_bpSaved          float64
l_bpFaced          float64
winner_hand_W_L    float64
winner_hand_W_R    float64
loser_hand_L_L     float64
loser_hand_L_R     float64
loser_hand_U       float64
Carpet             float64
Clay               float64
Grass              float64
Hard               float64
dtype: object


# Double check the shape of the dataframe after casting it and doing some operations on it (make sure that it's the same as before)
df.shape

(42923, 33)


# Creating a covariance matrix as a python data structure using scikit-learn
cols = df.columns
from sklearn.preprocessing import StandardScaler
stdsc = StandardScaler()
X_std = stdsc.fit_transform(df[cols].iloc[:,range(0,25)].values)
cov_mat = np.cov(X_std.T)


# Displaying the covariance matrix in a viewable format
sns.set_palette('pastel')

plt.figure(figsize=(20,20))
sns.set(font_scale=1.5)
hm = sns.heatmap(cov_mat,
                 cbar=True,
                 annot=True,
                 square=True,
                 fmt='.2f',
                 annot_kws={'size': 12},
                 yticklabels=cols,
                 xticklabels=cols)
plt.title('Covariance matrix showing correlation coefficients')
plt.tight_layout()
plt.show()


sns.set_palette('pastel')

# Create a grid to put our plots in
plt.rcParams["figure.autolayout"] = True

f, axes = plt.subplots(2, 4, figsize=(20, 8))

# Plot ages
ax = sns.boxplot(data=df[['winner_age', 'loser_age']], ax = axes[0][0], showfliers=False)
ax.set(title="Winners age to loser's age")

# Plot ranks
ax = sns.boxplot(data=df[['winner_rank', 'loser_rank']], ax = axes[0][1], showfliers=False)
ax.set(title="Winners rank to loser's rank")

# Plot 1st serve winning
ax = sns.boxplot(data=df[['w_1stWon', 'l_1stWon']], ax = axes[0][2], showfliers=False)
ax.set(title="Winner to loser 1st Serve winning %")

ax = sns.boxplot(data=df[['w_2ndWon', 'l_2ndWon']], ax = axes[0][3], showfliers=False)
ax.set(title="Winner to loser 2st Serve winning %")

# Post ace percentages
ax = sns.boxplot(data=df[['w_ace', 'l_ace']], ax = axes[1][0], showfliers=False)
ax.set(title="Winner to loser ace frequency")

# Plot break points faced
ax = sns.boxplot(data=df[['w_bpFaced', 'l_bpFaced']], ax = axes[1][1], showfliers=False)
ax.set(title="Winner to loser breakpoint frequency on serve")

# Plot break points saved proportions
winner_proportions = (df['w_bpSaved']/df['w_bpFaced'])
loser_proportions = (df['l_bpSaved']/df['l_bpFaced'])
data = {'Winner': winner_proportions, 'Loser': loser_proportions}
saved_df = pd.DataFrame(data)
ax = sns.boxplot(data=saved_df, ax=axes[1][2], showfliers=False)
ax.set(title="Winner to loser saving break points")

# Plot double fault frequencies
ax = sns.boxplot(data=df[['w_df', 'l_df']], ax = axes[1][3], showfliers=False)
ax.set(title="Winner to loser double fault frequency")

sns.displot(df[['winner_ht', 'loser_ht']], kind='kde').set(title='Winner to loser height distributions')

<seaborn.axisgrid.FacetGrid at 0x7f97a8a6b850>


df.columns

Index(['winner_ht', 'winner_age', 'winner_rank', 'loser_ht', 'loser_age',
       'loser_rank', 'w_ace', 'w_df', 'w_svpt', 'w_1stIn', 'w_1stWon',
       'w_2ndWon', 'w_SvGms', 'w_bpSaved', 'w_bpFaced', 'l_ace', 'l_df',
       'l_svpt', 'l_1stIn', 'l_1stWon', 'l_2ndWon', 'l_SvGms', 'l_bpSaved',
       'l_bpFaced', 'winner_hand_W_L', 'winner_hand_W_R', 'loser_hand_L_L',
       'loser_hand_L_R', 'loser_hand_U', 'Carpet', 'Clay', 'Grass', 'Hard'],
      dtype='object')


winner_proportions = ((df['w_1stWon']+df['w_2ndWon'])/(df['w_svpt']-df['w_df']))
loser_proportions = ((df['l_1stWon']+df['l_2ndWon'])/(df['l_svpt']-df['l_df']))
data = {'Winner % of won serve points': winner_proportions, 'Loser % of serve points': loser_proportions}
saved_df = pd.DataFrame(data)
ax = sns.scatterplot(data=saved_df)


saved_df.describe()


dfs = [df, saved_df]
df = pd.concat(dfs, axis = 1)
df


winner_proportions = (df['w_df']/df['w_svpt'])
loser_proportions = (df['l_df']/df['l_svpt'])
data = {'Winner Double Fault %': winner_proportions, 'Loser Double Fault %': loser_proportions}
df_df = pd.DataFrame(data)
ax = sns.scatterplot(data=df_df)


df_df.describe()


dfs = [df, df_df]
df = pd.concat(dfs, axis = 1)
df


# T-Test
# ANOVA F-Test

from sklearn.model_selection import train_test_split

#Use the train_test_split module to split our data

# retrieve numpy array
dataset = df.values

# split into input and output variables
x = dataset[:, 1:-2]
y = dataset[:,0].astype('int')

x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.2, random_state=42)

# summarize

print("# of training data points: %d" % x_train.shape[0])
print("# of testing data points: %d" % x_test.shape[0])
print("Class labels:", np.unique(y))

# of training data points: 34338
# of testing data points: 8585
Class labels: [163 168 170 172 173 175 178 180 181 183 185 188 190 193 194 196 198 201
 203 206 208]


df.columns

Index(['winner_ht', 'winner_age', 'winner_rank', 'loser_ht', 'loser_age',
       'loser_rank', 'w_ace', 'w_df', 'w_svpt', 'w_1stIn', 'w_1stWon',
       'w_2ndWon', 'w_SvGms', 'w_bpSaved', 'w_bpFaced', 'l_ace', 'l_df',
       'l_svpt', 'l_1stIn', 'l_1stWon', 'l_2ndWon', 'l_SvGms', 'l_bpSaved',
       'l_bpFaced', 'winner_hand_W_L', 'winner_hand_W_R', 'loser_hand_L_L',
       'loser_hand_L_R', 'loser_hand_U', 'Carpet', 'Clay', 'Grass', 'Hard',
       'Winner % of won serve points', 'Loser % of serve points',
       'Winner Double Fault %', 'Loser Double Fault %'],
      dtype='object')


df.isnull().sum().sum()/df.size
df = df.dropna()


import pandas as pd
from sklearn.model_selection import train_test_split

y = df['w_ace'] #RIGHT HERE
X = pd.DataFrame(df, columns = ['winner_ht', 'winner_age', 'winner_rank', 'w_df', 'w_svpt', 'w_1stIn', 'w_1stWon', 'w_2ndWon', 'w_SvGms', 'w_bpSaved',
       'w_bpFaced', 'random', 'Carpet', 'Clay', 'Grass', 'Hard',
       'Winner % of won serve points', 'Loser % of serve points',
       'Winner Double Fault %', 'Loser Double Fault %'])
np.random.seed(seed = 42)
X['random'] = np.random.random(size = len(X))
X_train, X_valid, y_train, y_valid = train_test_split(X, y, test_size = 0.3, random_state = 42)


X_train


from sklearn.ensemble import RandomForestRegressor

rf = RandomForestRegressor(n_estimators = 150,
                           n_jobs = -1,
                           oob_score = True,
                           bootstrap = True,
                           random_state = 42)
rf.fit(X_train, y_train)

print('R^2 Training Score: {:.2f} \nOOB Score: {:.2f} \nR^2 Validation Score: {:.2f}'.format(rf.score(X_train, y_train), 
                                                                                             rf.oob_score_,
                                                                                             rf.score(X_valid, y_valid)))

R^2 Training Score: 0.93 
OOB Score: 0.50 
R^2 Validation Score: 0.51


from sklearn.metrics import r2_score
from rfpimp import permutation_importances

def r2(rf, X_train, y_train):
    return r2_score(y_train, rf.predict(X_train))

perm_imp_rfpimp = permutation_importances(rf, X_train, y_train, r2)


perm_imp_rfpimp


perm_imp_rfpimp = perm_imp_rfpimp.reset_index()
perm_imp_rfpimp.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 20 entries, 0 to 19
Data columns (total 2 columns):
 #   Column      Non-Null Count  Dtype  
---  ------      --------------  -----  
 0   Feature     20 non-null     object 
 1   Importance  20 non-null     float64
dtypes: float64(1), object(1)
memory usage: 448.0+ bytes


# sns.set_palette('pastel')
# to_graph = pd.DataFrame(arr, columns=['Feature', 'Importance'])
# sns.set(rc={'figure.figsize':(15,15)})
# sns.barplot(x=to_graph['Feature'], y=to_graph['Importance'])
graph = sns.barplot(x=perm_imp_rfpimp['Feature'], y=perm_imp_rfpimp['Importance'])
sns.set(rc={'figure.figsize':(12,12)})
plt.xticks(rotation=45)
graph.tick_params(labelsize=8)


df.head()


from statsmodels.stats.outliers_influence import variance_inflation_factor


X = df.loc[:, df.columns != 'w_ace']
y = df['w_ace']


vif  = pd.DataFrame()
vif['VIF Factor'] = [variance_inflation_factor(X.values, i) for i in range(X.shape[1])]
vif["predictor"] = X.columns

/Users/mukunds/opt/anaconda3/lib/python3.9/site-packages/statsmodels/stats/outliers_influence.py:195: RuntimeWarning: divide by zero encountered in double_scalars
  vif = 1. / (1. - r_squared_i)
/Users/mukunds/opt/anaconda3/lib/python3.9/site-packages/statsmodels/regression/linear_model.py:1736: RuntimeWarning: invalid value encountered in double_scalars
  return 1 - self.ssr/self.centered_tss

vif


vif = vif[vif['VIF Factor'] <= 10]
vif


df.head()


from sklearn.linear_model import Ridge


# Perform OLS on the new dataframe
import statsmodels.api as sm
X_ols = df[vif['predictor']]
y_ols = y
X_ols = sm.add_constant(X_ols)
model_ols = sm.OLS(y_ols, X_ols).fit()
print(model_ols.summary())

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  w_ace   R-squared:                       0.369
Model:                            OLS   Adj. R-squared:                  0.369
Method:                 Least Squares   F-statistic:                     2091.
Date:                Fri, 12 May 2023   Prob (F-statistic):               0.00
Time:                        10:40:57   Log-Likelihood:            -1.2401e+05
No. Observations:               42913   AIC:                         2.481e+05
Df Residuals:                   42900   BIC:                         2.482e+05
Df Model:                          12                                         
Covariance Type:            nonrobust                                         
================================================================================================
                                   coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------------------------
const                          -40.8057      0.864    -47.236      0.000     -42.499     -39.112
winner_ht                        0.2519      0.003     77.462      0.000       0.245       0.258
winner_age                       0.0290      0.006      4.867      0.000       0.017       0.041
winner_rank                     -0.0019      0.000     -5.930      0.000      -0.003      -0.001
loser_ht                        -0.1130      0.003    -33.579      0.000      -0.120      -0.106
loser_age                        0.0480      0.006      8.206      0.000       0.037       0.060
loser_rank                       0.0008      0.000      3.770      0.000       0.000       0.001
l_ace                            0.1837      0.006     32.283      0.000       0.173       0.195
l_df                             0.7725      0.016     47.138      0.000       0.740       0.805
Winner % of won serve points    14.8815      0.335     44.483      0.000      14.226      15.537
Loser % of serve points         13.2404      0.326     40.559      0.000      12.601      13.880
Winner Double Fault %            9.6182      0.833     11.547      0.000       7.986      11.251
Loser Double Fault %           -59.1060      1.265    -46.726      0.000     -61.585     -56.627
==============================================================================
Omnibus:                    11059.616   Durbin-Watson:                   1.852
Prob(Omnibus):                  0.000   Jarque-Bera (JB):            49944.417
Skew:                           1.193   Prob(JB):                         0.00
Kurtosis:                       7.715   Cond. No.                     1.73e+04
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 1.73e+04. This might indicate that there are
strong multicollinearity or other numerical problems.


coef=[]
alphas = range(0,40)

for a in alphas:
  ridgereg = Ridge(alpha=a)
  ridgereg.fit(X, y)
  coef.append(ridgereg.coef_)

###Make plot of Beta as a function of Alpha
fig=plt.figure()
ax=fig.add_subplot(111)
ax.plot(alphas,coef)
ax.set_xlabel('Alpha (Regularization Parameter)')
ax.set_ylabel('Beta (Predictor Coefficients)')
ax.set_title('Ridge Coefficients vs Regularization Parameters')
ax.axis('tight')
plt.ylim(-1, 1)
plt.xlim(0, 35)
fig.savefig('coef_vs_alpha.png')


from sklearn.model_selection import train_test_split
from sklearn.metrics import r2_score

# Split data into training and testing sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# Create a Ridge regression object and fit the model to the training data
ridge = Ridge(alpha=35)
ridge.fit(X_train, y_train)

# Make predictions using the Ridge model on the testing data
y_pred = ridge.predict(X_test)

# Calculate the R-squared score
r2 = r2_score(y_test, y_pred)
print('R-squared score with Ridge regularization:', r2)

# Perform OLS on the new dataframe
import statsmodels.api as sm
X_ols = X
y_ols = y
X_ols = sm.add_constant(X_ols)
model_ols = sm.OLS(y_ols, X_ols).fit()
print(model_ols.summary())

R-squared score with Ridge regularization: 0.5227951439859293
                            OLS Regression Results                            
==============================================================================
Dep. Variable:                  w_ace   R-squared:                       0.526
Model:                            OLS   Adj. R-squared:                  0.526
Method:                 Least Squares   F-statistic:                     1486.
Date:                Fri, 12 May 2023   Prob (F-statistic):               0.00
Time:                        10:40:58   Log-Likelihood:            -1.1788e+05
No. Observations:               42913   AIC:                         2.358e+05
Df Residuals:                   42880   BIC:                         2.361e+05
Df Model:                          32                                         
Covariance Type:            nonrobust                                         
================================================================================================
                                   coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------------------------
const                           -8.2662      0.431    -19.192      0.000      -9.110      -7.422
winner_ht                        0.2009      0.003     69.884      0.000       0.195       0.207
winner_age                       0.0118      0.005      2.275      0.023       0.002       0.022
winner_rank                     -0.0006      0.000     -2.063      0.039      -0.001   -2.86e-05
loser_ht                        -0.0758      0.003    -25.638      0.000      -0.082      -0.070
loser_age                        0.0464      0.005      9.118      0.000       0.036       0.056
loser_rank                       0.0009      0.000      4.829      0.000       0.001       0.001
w_df                             0.1191      0.026      4.527      0.000       0.068       0.171
w_svpt                          -0.0861      0.009     -9.185      0.000      -0.104      -0.068
w_1stIn                         -0.3608      0.006    -57.871      0.000      -0.373      -0.349
w_1stWon                         0.7556      0.015     50.148      0.000       0.726       0.785
w_2ndWon                         0.2622      0.016     16.538      0.000       0.231       0.293
w_SvGms                         -0.0706      0.035     -1.992      0.046      -0.140      -0.001
w_bpSaved                       -0.3149      0.035     -9.005      0.000      -0.383      -0.246
w_bpFaced                        0.3260      0.033      9.835      0.000       0.261       0.391
l_ace                           -0.0091      0.006     -1.589      0.112      -0.020       0.002
l_df                             0.0730      0.022      3.353      0.001       0.030       0.116
l_svpt                          -0.1125      0.008    -14.007      0.000      -0.128      -0.097
l_1stIn                          0.0306      0.005      5.645      0.000       0.020       0.041
l_1stWon                         0.1573      0.015     10.769      0.000       0.129       0.186
l_2ndWon                         0.1807      0.015     12.028      0.000       0.151       0.210
l_SvGms                         -0.1196      0.036     -3.342      0.001      -0.190      -0.049
l_bpSaved                       -0.1498      0.036     -4.210      0.000      -0.220      -0.080
l_bpFaced                        0.1447      0.034      4.313      0.000       0.079       0.210
winner_hand_W_L                 -4.2999      0.218    -19.756      0.000      -4.726      -3.873
winner_hand_W_R                 -3.9663      0.217    -18.299      0.000      -4.391      -3.541
loser_hand_L_L                  -4.2301      0.218    -19.418      0.000      -4.657      -3.803
loser_hand_L_R                  -4.0361      0.216    -18.659      0.000      -4.460      -3.612
loser_hand_U                 -5.528e-16   5.56e-16     -0.994      0.320   -1.64e-15    5.38e-16
Carpet                          -1.3603      0.133    -10.225      0.000      -1.621      -1.100
Clay                            -3.3448      0.113    -29.503      0.000      -3.567      -3.123
Grass                           -1.6450      0.119    -13.803      0.000      -1.879      -1.411
Hard                            -1.9162      0.115    -16.723      0.000      -2.141      -1.692
Winner % of won serve points    -4.7771      0.705     -6.780      0.000      -6.158      -3.396
Loser % of serve points         -5.1020      0.593     -8.603      0.000      -6.264      -3.940
Winner Double Fault %           -1.0228      1.972     -0.519      0.604      -4.888       2.843
Loser Double Fault %             1.7799      1.536      1.159      0.247      -1.231       4.791
==============================================================================
Omnibus:                     5296.377   Durbin-Watson:                   1.949
Prob(Omnibus):                  0.000   Jarque-Bera (JB):            12961.784
Skew:                           0.719   Prob(JB):                         0.00
Kurtosis:                       5.276   Cond. No.                     1.56e+17
==============================================================================

Notes:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The smallest eigenvalue is 1.8e-25. This might indicate that there are
strong multicollinearity problems or that the design matrix is singular.


s1 = model_ols.predict(X_ols)
s2 = df['w_ace']
s3 = 1 - abs(s2 - s1)/s2
data = {'Actual': s2, 'Predicted': s1, 'Accuracy': s3}
df_to_graph = pd.DataFrame(data)


df_to_graph


df_g = df_to_graph[df_to_graph['Accuracy'] != float('inf')]
sns.boxplot(y=df_g['Accuracy'], showfliers=False)

<AxesSubplot:ylabel='Accuracy'>


sns.lineplot(data=df_to_graph[['Predicted', 'Actual']])

<AxesSubplot:>

	tourney_id	tourney_name	surface	draw_size	tourney_level	tourney_date	match_num	winner_id	winner_seed	winner_entry	...	w_bpFaced	l_ace	l_df	l_svpt	l_1stIn	l_1stWon	l_2ndWon	l_SvGms	l_bpSaved	l_bpFaced
0	2016-M020	Brisbane	Hard	32.0	A	20160104.0	300.0	105683.0	4.0	NaN	...	1.0	7.0	3.0	61.0	34.0	25.0	14.0	10.0	3.0	5.0
1	2016-M020	Brisbane	Hard	32.0	A	20160104.0	299.0	103819.0	1.0	NaN	...	1.0	2.0	4.0	55.0	31.0	18.0	9.0	8.0	2.0	6.0
2	2016-M020	Brisbane	Hard	32.0	A	20160104.0	298.0	105683.0	4.0	NaN	...	4.0	10.0	3.0	84.0	54.0	41.0	16.0	12.0	2.0	2.0
3	2016-M020	Brisbane	Hard	32.0	A	20160104.0	297.0	103819.0	1.0	NaN	...	1.0	8.0	2.0	104.0	62.0	46.0	21.0	16.0	8.0	11.0
4	2016-M020	Brisbane	Hard	32.0	A	20160104.0	296.0	106233.0	8.0	NaN	...	2.0	11.0	2.0	98.0	52.0	41.0	27.0	15.0	7.0	8.0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
2999	2016-M-DC-2016-WG-PO-KAZ-RUS-01	Davis Cup WG PO: KAZ vs RUS	Hard	4.0	D	20160916.0	4.0	105723.0	NaN	NaN	...	1.0	2.0	2.0	87.0	60.0	34.0	8.0	11.0	7.0	13.0
3000	2016-M-DC-2016-WG-PO-SUI-UZB-01	Davis Cup WG PO: SUI vs UZB	Clay	4.0	D	20160916.0	1.0	104797.0	NaN	NaN	...	1.0	4.0	3.0	88.0	53.0	35.0	14.0	13.0	4.0	8.0
3001	2016-M-DC-2016-WG-PO-SUI-UZB-01	Davis Cup WG PO: SUI vs UZB	Clay	4.0	D	20160916.0	2.0	105967.0	NaN	NaN	...	3.0	3.0	1.0	107.0	73.0	36.0	18.0	12.0	10.0	16.0
3002	2016-M-DC-2016-WG-PO-SUI-UZB-01	Davis Cup WG PO: SUI vs UZB	Clay	4.0	D	20160916.0	4.0	104797.0	NaN	NaN	...	8.0	3.0	6.0	169.0	103.0	70.0	34.0	24.0	2.0	7.0
3003	2016-M-DC-2016-WG-PO-SUI-UZB-01	Davis Cup WG PO: SUI vs UZB	Clay	4.0	D	20160916.0	5.0	134472.0	NaN	NaN	...	8.0	8.0	3.0	114.0	63.0	39.0	26.0	19.0	6.0	14.0

	tourney_id	tourney_name	surface	draw_size	tourney_level	tourney_date	match_num	winner_id	winner_seed	winner_entry	...	w_bpFaced	l_ace	l_df	l_svpt	l_1stIn	l_1stWon	l_2ndWon	l_SvGms	l_bpSaved	l_bpFaced
0	2017-M020	Brisbane	Hard	32	A	20170102	300	105777	7.0	NaN	...	7.0	4.0	0.0	69.0	49.0	36.0	9.0	12.0	2.0	5.0
1	2017-M020	Brisbane	Hard	32	A	20170102	299	105777	7.0	NaN	...	0.0	4.0	3.0	61.0	28.0	24.0	16.0	10.0	2.0	4.0
2	2017-M020	Brisbane	Hard	32	A	20170102	298	105453	3.0	NaN	...	5.0	9.0	2.0	61.0	37.0	27.0	10.0	10.0	0.0	2.0
3	2017-M020	Brisbane	Hard	32	A	20170102	297	105683	1.0	NaN	...	7.0	4.0	0.0	84.0	61.0	39.0	14.0	14.0	2.0	4.0
4	2017-M020	Brisbane	Hard	32	A	20170102	296	105777	7.0	NaN	...	14.0	6.0	5.0	82.0	37.0	29.0	24.0	14.0	4.0	7.0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
383	2017-M-DC-2017-WG-M-RUS-SRB-01	Davis Cup WG R1: RUS vs SRB	Hard	4	D	20170203	5	106421	NaN	NaN	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
384	2017-M-DC-2017-WG-M-SUI-USA-01	Davis Cup WG R1: SUI vs USA	Hard	4	D	20170203	1	106058	NaN	NaN	...	4.0	4.0	5.0	82.0	52.0	38.0	9.0	13.0	7.0	12.0
385	2017-M-DC-2017-WG-M-SUI-USA-01	Davis Cup WG R1: SUI vs USA	Hard	4	D	20170203	2	104545	NaN	NaN	...	9.0	7.0	9.0	113.0	76.0	53.0	17.0	19.0	7.0	11.0
386	2017-M-DC-2017-WG-M-SUI-USA-01	Davis Cup WG R1: SUI vs USA	Hard	4	D	20170203	4	105023	NaN	NaN	...	0.0	3.0	2.0	76.0	42.0	30.0	14.0	10.0	8.0	10.0
387	2017-M-DC-2017-WG-M-SUI-USA-01	Davis Cup WG R1: SUI vs USA	Hard	4	D	20170203	5	105449	NaN	NaN	...	4.0	1.0	2.0	54.0	27.0	15.0	15.0	9.0	2.0	6.0

	tourney_id	tourney_name	surface	draw_size	tourney_level	tourney_date	match_num	winner_id	winner_seed	winner_entry	...	w_bpFaced	l_ace	l_df	l_svpt	l_1stIn	l_1stWon	l_2ndWon	l_SvGms	l_bpSaved	l_bpFaced
0	2000-717	Orlando	Clay	32.0	A	20000501.0	1.0	102179.0	NaN	NaN	...	15.0	13.0	4.0	110.0	59.0	49.0	31.0	17.0	4.0	4.0
1	2000-717	Orlando	Clay	32.0	A	20000501.0	2.0	103602.0	NaN	Q	...	6.0	0.0	0.0	57.0	24.0	13.0	17.0	10.0	4.0	9.0
2	2000-717	Orlando	Clay	32.0	A	20000501.0	3.0	103387.0	NaN	NaN	...	0.0	2.0	2.0	65.0	39.0	22.0	10.0	8.0	6.0	10.0
3	2000-717	Orlando	Clay	32.0	A	20000501.0	4.0	101733.0	NaN	NaN	...	12.0	4.0	6.0	104.0	57.0	35.0	24.0	15.0	6.0	11.0
4	2000-717	Orlando	Clay	32.0	A	20000501.0	5.0	101727.0	4.0	NaN	...	1.0	0.0	3.0	47.0	28.0	17.0	10.0	8.0	3.0	6.0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
383	2017-M-DC-2017-WG-M-RUS-SRB-01	Davis Cup WG R1: RUS vs SRB	Hard	4.0	D	20170203.0	5.0	106421.0	NaN	NaN	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
384	2017-M-DC-2017-WG-M-SUI-USA-01	Davis Cup WG R1: SUI vs USA	Hard	4.0	D	20170203.0	1.0	106058.0	NaN	NaN	...	4.0	4.0	5.0	82.0	52.0	38.0	9.0	13.0	7.0	12.0
385	2017-M-DC-2017-WG-M-SUI-USA-01	Davis Cup WG R1: SUI vs USA	Hard	4.0	D	20170203.0	2.0	104545.0	NaN	NaN	...	9.0	7.0	9.0	113.0	76.0	53.0	17.0	19.0	7.0	11.0
386	2017-M-DC-2017-WG-M-SUI-USA-01	Davis Cup WG R1: SUI vs USA	Hard	4.0	D	20170203.0	4.0	105023.0	NaN	NaN	...	0.0	3.0	2.0	76.0	42.0	30.0	14.0	10.0	8.0	10.0
387	2017-M-DC-2017-WG-M-SUI-USA-01	Davis Cup WG R1: SUI vs USA	Hard	4.0	D	20170203.0	5.0	105449.0	NaN	NaN	...	4.0	1.0	2.0	54.0	27.0	15.0	15.0	9.0	2.0	6.0

	surface	winner_name	winner_hand	winner_ht	winner_age	winner_rank	loser_name	loser_hand	loser_ht	loser_age	...	w_bpFaced	l_ace	l_df	l_svpt	l_1stIn	l_1stWon	l_2ndWon	l_SvGms	l_bpSaved	l_bpFaced
0	Clay	Antony Dupuis	R	185.0	27.181383	113.0	Andrew Ilie	R	180.0	24.035592	...	15.0	13.0	4.0	110.0	59.0	49.0	31.0	17.0	4.0	4.0
1	Clay	Fernando Gonzalez	R	183.0	19.756331	352.0	Cecil Mamiit	R	173.0	23.843943	...	6.0	0.0	0.0	57.0	24.0	13.0	17.0	10.0	4.0	9.0
2	Clay	Paradorn Srichaphan	R	185.0	20.881588	103.0	Sebastien Lareau	R	183.0	27.011636	...	0.0	2.0	2.0	65.0	39.0	22.0	10.0	8.0	6.0	10.0
3	Clay	Jan Siemerink	L	183.0	30.047912	107.0	Justin Gimelstob	R	196.0	23.260780	...	12.0	4.0	6.0	104.0	57.0	35.0	24.0	15.0	6.0	11.0
4	Clay	Jason Stoltenberg	R	185.0	30.075291	74.0	Alex Lopez Moron	R	175.0	29.423682	...	1.0	0.0	3.0	47.0	28.0	17.0	10.0	8.0	3.0	6.0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
383	Hard	Daniil Medvedev	R	NaN	20.969199	63.0	Nenad Zimonjic	R	190.0	40.657084	...	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
384	Hard	Jack Sock	R	185.0	24.350445	20.0	Marco Chiudinelli	R	185.0	35.389459	...	4.0	4.0	5.0	82.0	52.0	38.0	9.0	13.0	7.0	12.0
385	Hard	John Isner	R	206.0	31.764545	23.0	Henri Laaksonen	R	NaN	24.835044	...	9.0	7.0	9.0	113.0	76.0	53.0	17.0	19.0	7.0	11.0
386	Hard	Sam Querrey	R	198.0	29.316906	27.0	Adrien Bossel	L	NaN	30.091718	...	0.0	3.0	2.0	76.0	42.0	30.0	14.0	10.0	8.0	10.0
387	Hard	Steve Johnson	R	188.0	27.101985	31.0	Antoine Bellier	U	NaN	20.284736	...	4.0	1.0	2.0	54.0	27.0	15.0	15.0	9.0	2.0	6.0

	surface	winner_name	winner_hand	winner_ht	winner_age	winner_rank	loser_name	loser_hand	loser_ht	loser_age	...	l_1stWon	l_2ndWon	l_SvGms	l_bpSaved	l_bpFaced	winner_hand_W_L	winner_hand_W_R	loser_hand_L_L	loser_hand_L_R	loser_hand_U
0	Clay	Antony Dupuis	W_R	185.0	27.181383	113.0	Andrew Ilie	L_R	180.0	24.035592	...	49.0	31.0	17.0	4.0	4.0	0	1	0	1	0
1	Clay	Fernando Gonzalez	W_R	183.0	19.756331	352.0	Cecil Mamiit	L_R	173.0	23.843943	...	13.0	17.0	10.0	4.0	9.0	0	1	0	1	0
2	Clay	Paradorn Srichaphan	W_R	185.0	20.881588	103.0	Sebastien Lareau	L_R	183.0	27.011636	...	22.0	10.0	8.0	6.0	10.0	0	1	0	1	0
3	Clay	Jan Siemerink	W_L	183.0	30.047912	107.0	Justin Gimelstob	L_R	196.0	23.260780	...	35.0	24.0	15.0	6.0	11.0	1	0	0	1	0
4	Clay	Jason Stoltenberg	W_R	185.0	30.075291	74.0	Alex Lopez Moron	L_R	175.0	29.423682	...	17.0	10.0	8.0	3.0	6.0	0	1	0	1	0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
358	Clay	Andreas Seppi	W_R	190.0	32.941821	68.0	Carlos Berlocq	L_R	183.0	33.990418	...	40.0	22.0	17.0	10.0	17.0	0	1	0	1	0
359	Clay	Carlos Berlocq	W_R	183.0	33.990418	81.0	Paolo Lorenzi	L_R	183.0	35.126626	...	63.0	24.0	22.0	9.0	15.0	0	1	0	1	0
361	Hard	Steve Darcis	W_R	178.0	32.884326	58.0	Philipp Kohlschreiber	L_R	178.0	33.292266	...	82.0	24.0	25.0	6.0	11.0	0	1	0	1	0
367	Hard	Vasek Pospisil	W_R	193.0	26.606434	133.0	Daniel Evans	L_R	175.0	26.691307	...	67.0	25.0	21.0	5.0	10.0	0	1	0	1	0
384	Hard	Jack Sock	W_R	185.0	24.350445	20.0	Marco Chiudinelli	L_R	185.0	35.389459	...	38.0	9.0	13.0	7.0	12.0	0	1	0	1	0

Which factors affect the winners of a tennis match the most?

Data Collection

Exploratory Analysis

Data Visualization

Hypothesis Testing¶

Machine Learning

In the following section, we are flattening data to combat multilinearity.¶

OLS on our dataset¶

Using OLS to predict (after flattening data through ridge regularization)¶

	Winner % of won serve points	Loser % of serve points
count	42916.000000	42915.000000
mean	0.711841	0.598190
std	0.068678	0.078835
min	0.323944	0.000000
25%	0.663717	0.551020
50%	0.708738	0.601266
75%	0.756098	0.649658
max	1.000000	1.000000

	winner_ht	winner_age	winner_rank	w_df	w_svpt	w_1stIn	w_1stWon	w_2ndWon	w_SvGms	w_bpSaved	w_bpFaced	random	Carpet	Clay	Grass	Hard	Winner % of won serve points	Loser % of serve points	Winner Double Fault %	Loser Double Fault %
2029	190.0	23.310062	137.0	5.0	94.0	52.0	32.0	16.0	14.0	4.0	10.0	0.516848	0.0	0.0	0.0	1.0	0.539326	0.550000	0.053191	0.056604
471	183.0	21.240246	142.0	0.0	56.0	32.0	21.0	17.0	10.0	0.0	1.0	0.359999	0.0	0.0	0.0	1.0	0.678571	0.639344	0.000000	0.115942
2590	183.0	28.073922	22.0	3.0	115.0	71.0	53.0	25.0	17.0	6.0	8.0	0.482316	0.0	0.0	0.0	1.0	0.696429	0.681034	0.026087	0.016949
207	185.0	24.796715	54.0	3.0	110.0	55.0	43.0	26.0	12.0	9.0	12.0	0.578141	1.0	0.0	0.0	0.0	0.644860	0.620690	0.027273	0.000000
1392	180.0	32.501027	66.0	2.0	125.0	65.0	49.0	37.0	18.0	6.0	7.0	0.013986	0.0	0.0	0.0	1.0	0.699187	0.642857	0.016000	0.017544
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
700	185.0	22.754278	44.0	1.0	67.0	47.0	31.0	10.0	9.0	6.0	7.0	0.500877	0.0	0.0	0.0	1.0	0.621212	0.535714	0.014925	0.096774
292	180.0	34.491444	7.0	1.0	83.0	55.0	40.0	16.0	12.0	5.0	6.0	0.879721	0.0	0.0	0.0	1.0	0.682927	0.640449	0.012048	0.032609
857	183.0	29.045859	3.0	1.0	34.0	23.0	18.0	10.0	7.0	0.0	0.0	0.393821	0.0	1.0	0.0	0.0	0.848485	0.428571	0.029412	0.027778
914	175.0	21.697467	78.0	4.0	79.0	48.0	30.0	16.0	10.0	8.0	11.0	0.775528	0.0	1.0	0.0	0.0	0.613333	0.533333	0.050633	0.013158
2158	188.0	22.461328	3.0	1.0	78.0	55.0	44.0	16.0	12.0	4.0	4.0	0.210462	0.0	0.0	0.0	1.0	0.779221	0.684211	0.012821	0.040404

	Importance
Feature
w_1stWon	0.750639
winner_ht	0.441957
Winner % of won serve points	0.397995
Clay	0.163610
w_1stIn	0.157716
w_2ndWon	0.106801
Loser % of serve points	0.106719
w_svpt	0.103256
winner_age	0.061434
winner_rank	0.060554
random	0.057177
Loser Double Fault %	0.052967
Winner Double Fault %	0.051709
w_SvGms	0.045749
w_bpFaced	0.024362
w_bpSaved	0.022036
w_df	0.011356
Hard	0.005502
Grass	0.003580
Carpet	0.001827

	VIF Factor	predictor
0	1.134959	winner_ht
1	1.031203	winner_age
2	1.044735	winner_rank
3	1.208028	loser_ht
4	1.028227	loser_age
5	1.034275	loser_rank
6	10.739158	w_df
7	224.615461	w_svpt
8	41.329145	w_1stIn
9	123.998059	w_1stWon
10	37.977988	w_2ndWon
11	67.240580	w_SvGms
12	34.908152	w_bpSaved
13	53.831384	w_bpFaced
14	2.270725	l_ace
15	9.086078	l_df
16	165.876667	l_svpt
17	32.463603	l_1stIn
18	131.562031	l_1stWon
19	36.110525	l_2ndWon
20	68.992944	l_SvGms
21	40.583832	l_bpSaved
22	57.987006	l_bpFaced
23	inf	winner_hand_W_L
24	inf	winner_hand_W_R
25	inf	loser_hand_L_L
26	inf	loser_hand_L_R
27	NaN	loser_hand_U
28	inf	Carpet
29	inf	Clay
30	inf	Grass
31	inf	Hard
32	7.051684	Winner % of won serve points
33	6.577546	Loser % of serve points
34	7.707932	Winner Double Fault %
35	6.841490	Loser Double Fault %

	Actual	Predicted	Accuracy
0	8.0	10.261121	0.717360
1	4.0	3.436732	0.859183
2	4.0	3.335219	0.833805
3	8.0	5.752750	0.719094
4	3.0	5.298179	0.233940
...	...	...	...
358	8.0	8.968136	0.878983
359	4.0	7.172974	0.206756
361	4.0	7.975373	0.006157
367	25.0	16.446268	0.657851
384	12.0	7.769555	0.647463